Benchmark Almanac

DAYJOB: Finance

Claude Opus 5.5 42.5% leads since September 2026
In use

Agents get short requests from finance colleagues plus a workspace of filings, ledgers and emails. An agent judge applies an expert rubric, and a pass needs 95% of criteria.

80 tasks. Surge calls DAYJOB its evolution of GDPval. Also hosted on Kaggle Benchmarks and in the Fireworks Specialized Intelligence Index.

Made by
Surge AI
Released
September 2026
Kind
Tasks from real jobs
Level
Expert
Scored
Graded by a model
Answers
Agent task
Tasks
All public
Strongest model tested
GPT-6 Astra #3 on the capability index
Last in a launch post
November 2026
Last model added
November 2026
Scores refreshed
November 12th, 2026
40
Entries on its own leaderboard, 39 models we track
1
Launch post lists it, from 1 lab

Claude Opus 5.5 leads with 42.5%

Best score per entry from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.

40 entries. Scroll sideways for the rest.

  1. 42.5 Claude Opus 5.5
  2. 40.5 Claude Sonnet 5.5
  3. 38.8 Muse Wonder 1.0
  4. 37.3 GPT-6 Astra
  5. 36.5 Claude Fable 5.1
  6. 33.5 GPT-6.1 Sol
  7. 30.5 Gemini 4 Argon
  8. 27.0 Muse Spark 1.3
  9. 26.5 Grok 4.7
  10. 23.8 Claude Opus 5
  11. 21.0 Grok 4.6
  12. 19.5 GPT-6 Sol
  13. 18.3 Claude Fable 5
  14. 17.0 Kimi K3.1 Pro
  15. 14.3 Inkling 2 Small
  16. 12.5 DeepSeek V4.1 Flash 1017
  17. 11.8 GPT-5.6 Sol
  18. 11.0 GLM-5.3
  19. 9.30 Tencent Hy4 preview
  20. 8.80 Qwen 3.8 Max
  21. 7.30 GLM-5.3-Flash
  22. 6.80 Muse Spark 1.2
  23. 6.80 GPT-6 Luna
  24. 5.50 Kimi K3
  25. 5.30 Gemini 3.7 Flash
  26. 4.80 Mistral Large 4
  27. 4.50 Claude Sonnet 5
  28. 4.50 Gemini 3.8 Flash
  29. 4.30 GPT-5.6 Luna
  30. 3.50 DeepSeek-V4-Pro
  31. 3.30 GPT-5.6 Terra
  32. 1.50 Beam 1.0
  33. 0.50 Tencent Hy3
  34. 0.30 Gemini 3.1 Pro
  35. 0.30 DeepSeek-V4-Flash
  36. 0.0 Mistral Large 3
  37. 0.0 Nemotron 3 Ultra
  38. 0.0 Kimi K2.7 Code
  39. 0.0 Inkling
  40. 0.0 Muse Glimmer
  1. 1 Claude Opus 5.5 adaptive/max 42.5%
  2. 2 Claude Sonnet 5.5 adaptive/max 40.5%
  3. 3 Muse Wonder 1.0 max 38.8%
  4. 4 GPT-6 Astra max 37.3%
  5. 5 Claude Fable 5.1 adaptive/max 36.5%
  6. 6 GPT-6.1 Sol max 33.5%
  7. 7 Gemini 4 Argon high 30.5%
  8. 8 Muse Spark 1.3 max 27%
  9. 9 Grok 4.7 xhigh 26.5%
  10. 10 Claude Opus 5 adaptive/max 23.8%
  11. 11 Grok 4.6 xhigh 21%
  12. 12 GPT-6 Sol max 19.5%
  13. 13 Claude Fable 5 adaptive/max 18.3%
  14. 14 Kimi K3.1 Pro max 17%
  15. 15 Inkling 2 Small xhigh 14.3%
  16. 16 DeepSeek V4.1 Flash 1017 max 12.5%
  17. 17 GPT-5.6 Sol max 11.8%
  18. 18 GLM-5.3 max 11%
  19. 19 Tencent Hy4 preview high 9.30%
  20. 20 Qwen 3.8 Max xhigh 8.80%
  21. 21 GLM-5.3-Flash max 7.30%
  22. 22 Muse Spark 1.2 xhigh 6.80%
  23. 23 GPT-6 Luna max 6.80%
  24. 24 Kimi K3 max 5.50%
  25. 25 Gemini 3.7 Flash high 5.30%
  26. 26 Mistral Large 4 high 4.80%
  27. 27 Claude Sonnet 5 adaptive/max 4.50%
  28. 28 Gemini 3.8 Flash high 4.50%
  29. 29 GPT-5.6 Luna max 4.30%
  30. 30 DeepSeek-V4-Pro max 3.50%
  31. 31 GPT-5.6 Terra xhigh 3.30%
  32. 32 Beam 1.0 max 1.50%
  33. 33 Tencent Hy3 high 0.50%
  34. 34 Gemini 3.1 Pro high 0.30%
  35. 35 DeepSeek-V4-Flash max 0.30%
  36. 36 Mistral Large 3 0%
  37. 37 Nemotron 3 Ultra 0%
  38. 38 Kimi K2.7 Code max 0%
  39. 39 Inkling max 0%
  40. 40 Muse Glimmer xhigh 0%

Source: Official DAYJOB: Finance leaderboard

#EntryScoreSettingDateSource
1 Claude Opus 5.5 42.5% adaptive/max Sep 2026 Official leaderboard
2 Claude Sonnet 5.5 40.5% adaptive/max Sep 2026 Official leaderboard
3 Muse Wonder 1.0 38.8% max Official leaderboard
4 GPT-6 Astra 37.3% max Sep 2026 Official leaderboard
5 Claude Fable 5.1 36.5% adaptive/max Sep 2026 Official leaderboard
6 GPT-6.1 Sol 33.5% max Sep 2026 Official leaderboard
7 Gemini 4 Argon 30.5% high Sep 2026 Official leaderboard
8 Muse Spark 1.3 27% max Sep 2026 Official leaderboard
9 Grok 4.7 26.5% xhigh Sep 2026 Official leaderboard
10 Claude Opus 5 23.8% adaptive/max Jul 2026 Official leaderboard
11 Grok 4.6 21% xhigh Aug 2026 Official leaderboard
12 GPT-6 Sol 19.5% max Sep 2026 Official leaderboard
13 Claude Fable 5 18.3% adaptive/max Jun 2026 Official leaderboard
14 Kimi K3.1 Pro 17% max Oct 2026 Official leaderboard
15 Inkling 2 Small 14.3% xhigh Nov 2026 Official leaderboard
16 DeepSeek V4.1 Flash 1017 12.5% max Oct 2026 Official leaderboard
17 GPT-5.6 Sol 11.8% max Jul 2026 Official leaderboard
18 GLM-5.3 11% max Aug 2026 Official leaderboard
19 Tencent Hy4 preview 9.30% high Aug 2026 Official leaderboard
20 Qwen 3.8 Max 8.80% xhigh Aug 2026 Official leaderboard
21 GLM-5.3-Flash 7.30% max Aug 2026 Official leaderboard
22 Muse Spark 1.2 6.80% xhigh Aug 2026 Official leaderboard
23 GPT-6 Luna 6.80% max Sep 2026 Official leaderboard
24 Kimi K3 5.50% max Jul 2026 Official leaderboard
25 Gemini 3.7 Flash 5.30% high Aug 2026 Official leaderboard
26 Mistral Large 4 4.80% high Oct 2026 Official leaderboard
27 Claude Sonnet 5 4.50% adaptive/max Jun 2026 Official leaderboard
28 Gemini 3.8 Flash 4.50% high Sep 2026 Official leaderboard
29 GPT-5.6 Luna 4.30% max Jul 2026 Official leaderboard
30 DeepSeek-V4-Pro 3.50% max Apr 2026 Official leaderboard
31 GPT-5.6 Terra 3.30% xhigh Jul 2026 Official leaderboard
32 Beam 1.0 1.50% max Oct 2026 Official leaderboard
33 Tencent Hy3 0.50% high Jul 2026 Official leaderboard
34 Gemini 3.1 Pro 0.30% high Feb 2026 Official leaderboard
35 DeepSeek-V4-Flash 0.30% max Apr 2026 Official leaderboard
36 Mistral Large 3 0% Dec 2025 Official leaderboard
37 Nemotron 3 Ultra 0% Jun 2026 Official leaderboard
38 Kimi K2.7 Code 0% max Jun 2026 Official leaderboard
39 Inkling 0% max Jul 2026 Official leaderboard
40 Muse Glimmer 0% xhigh Aug 2026 Official leaderboard

Labs report scores for 1 model in their cards. We plot every source in the charts below.

3 models have held number one on the official leaderboard since its release in September 2026

Lab claims count from the day the card came out, independent runs from the model’s release.

Anthropic OpenAI
Own leaderboard 3 leaders
Claude Opus 5.5 42.5%

Source: Official DAYJOB: Finance leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Claude Opus 5.5 GPT-6 Astra Claude Opus 5 Claude Fable 5 DeepSeek-V4-Pro Mistral Large 3 0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Claude Opus 5.5 Claude Opus 5

Sources: Official DAYJOB: Finance leaderboard and labs’ model cards

One lab put it in a launch post in the last six months

Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.

Thinking Machines
DateLaunch postScoreAs the post writes it
Nov 5, 2026 Inkling 2 Small 14.3 DAYJOB (Finance)

One lab claim next to the leaderboard

Each score a lab reported for its own model, beside the same model's best score on the leaderboard. Gaps mostly come from different harnesses, effort levels and task subsets. 1 of them match, which usually means one side copied the other. Every benchmark.

Model Lab says Board says Gap
Inkling 2 Small 14.3% DAYJOB (Finance) 14.3% Leaderboard, xhigh Same

More in Economic work

All 97 in Economic work
  • JobBench In use 130 tasks from 35 occupations workers most want delegated. Agent works through messy reference files, graded by fact-anchored binary rubrics. GLM-5.3 61.4%
  • GDPval In use Real work tasks from 44 occupations with reference files. Experts blindly compare model deliverables with expert work. Metric is win-or-tie rate. Seed2.1 Pro 87.9%
  • GDPval-AA v2.1 In use Agents with shell and web access complete 220 GDPval gold tasks from 44 occupations. Blind pairwise comparisons give Elo ratings. Claude Opus 5.5 1,846
  • AA-Briefcase v1.1 In use Version 1.1 of Artificial Analysis's multi-hour office work evaluation, with Elo ratings fitted by Crowd-BT. Claude Opus 5.5 1,822
  • CoWorkBench In use Agent completes long-horizon productivity tasks across computer science, finance, law and medicine. Qwen in-house scoring (method unpublished). Claude Fable 5 75.9%
  • OfficeQA In use Grounded reasoning over a large corpus of U.S. Treasury Bulletin documents: locate tables and compute precise numeric answers. Claude Opus 4.7 86.3%